Simplified Stack Research Articles

Survival Stacking Ensemble Model for Lung Cancer Risk Prediction.

The most well-established risk factor for lung cancer (LC) is smoking, responsible for approximately 85% of cases. The Lung Cancer Risk Assessment Tool (LCRAT) is a key advancement in this field, which predicts individual risk based on factors like smoking habits, demographic details, personal and family medical history, and environmental exposures. This paper proposes a model with fewer features that improves state of the art performance, using a simplified stacking ensemble, making it more accessible and easier to implement in routine healthcare practice. The data used in this work were derived from two cohorts in the United States: The National Lung Screening Trial (NLST) and the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial. Both our model and LCRAT achieve an AUC of 0.799 and 0.782 on test respectively. In terms of percentage of positives, in the 50% of the population, both detect 0.766 and 0.754 of the cases. The ensemble of different survival models enhances robustness by mitigating the weakness of individual models and directly impacts the efficiency of the model, increasing the efficiency and generalizability.

SStackGNN: Graph Data Augmentation Simplified Stacking Graph Neural Network for Twitter Bot Detection

Ensemble learning has the potential to enhance the efficacy of feeble classifiers significantly and is increasingly being utilized in Twitter bot detection. Previous methods have utilized stacking techniques to train the primary classifiers, implementing cross-validation to mitigate overfitting and enhance predictive accuracy. However, cross-validation substantially amplifies the computation time associated with stacking. To overcome this challenge, this paper presents a novel approach, the Simplified Stacking Graph Neural Network (SStackGNN), specifically designed for the detection of social bots. Our methodology leverages the power of Graph Neural Networks (GNNs) as base classifiers, enabling effective capturing of inter-account interactions. In addition, a Multilayer Perceptron (MLP) serves as a secondary classifier, amalgamating the outcomes of the foundational classifiers to generate final predictions, thereby enhancing the predictive performance of the GNNs. Instead of relying on cross-validation and distinct base classifier structures, we employ node-level, edge-level, and feature-level graph data augmentation techniques to acquire diverse foundational classifiers. This approach significantly alleviates the computational complexity while achieving superior performance. Experimental results demonstrate that our proposed SStackGNN outperforms other approaches.

Low velocity impact response of non-traditional double-double laminates

A novel family of composite laminates with a simplified stacking sequence and double-double layup [±ϕ/±ψ] n has significant potential to reduce weight and increase strength, while facilitating design optimization and simplifying the manufacturing process. In this study, the low-velocity impact response of a double-double (DD) laminate and a quadriaxial (Quad) laminate of equivalent stiffness and thickness were compared. Carbon/epoxy laminates were produced with stacking sequences of [±0/±50]10 and [03/90/±45/02/±45]2S corresponding to double-double and quadriaxial varieties, respectively. Low velocity impact tests were conducted at 74 J of energy and damage areas were examined using X-ray computed tomography. Compressive strength and compression after impact (CAI) strength were equivalent for the two laminates. However, differences in impact damage morphology were observed and are discussed.

Improved Two-Layer Stacking Model for Prediction of the Level of Delay Caused by Crashes: An Empirical Analysis of Texas

Road crashes cause significant traffic delay, which can bring unnecessary financial losses. The objective of this study is to predict the level of delay caused by crashes (LDC) and discuss significant risk factors. To ensure the efficiency and accuracy of prediction, an improved stacking model was developed using Texas crash data of 2020. The first layer integrates seven base classifiers and the second layer tests three classifiers with different advantages. To improve and simplify the stacking model, three state-of-the-art methods—Bayesian hyperparameter optimization (BO), multiobjective feature selection (FS), and ensemble selection (ES)—were used. First, the hyperparameters and the least and most effective features were selected for each base classifier by BO and FS, respectively. Then ES, considering diversity and performance, selects the least base classifiers to reduce the input of the second layer. Finally, permutation feature importance was used to interpret the best stacking model. The results indicate that the stacking model achieves superior performance on four indicators: recall, G mean, F1 score, and area under the receiver operating characteristic (ROC) curve (AUC-ROC). FS significantly improves the efficiency of the stacking model and ES obtains a simplified stacking model without significantly reducing performance. In addition, the combination of the two methods (FS and ES) tends to achieve the best performance, and six risk factors have the greatest contributions in prediction using permutation feature importance. The prediction of LDC and the analysis of the main contributing factors help road managers respond to the rescue strategies to mitigate traffic congestion caused by crashes in a timely manner, thus minimizing economic losses.

Numerical simulation acceleration of flat-chip solid oxide cell stacks by data-driven surrogate cell submodels

Three-dimensional (3D) multiphysics models are powerful tools for investigating the distributions of physical quantities such as temperature inside solid oxide cell (SOC) stacks, but their high computational cost remains an obstacle to their application in simulating industrial-scale stacks with tens of cells. To accelerate the simulation for the 3D model of a novel flat-chip SOC (FCSOC) stack, this study proposes a simplification method that replaces part of the governing equations with data-driven surrogate cell submodels. The submodels, built with the adaptive polynomial approximation (APA) method, take the form of polynomials and are easy to integrate into commercial CFD software such as COMSOL. Simulation shows that the simplified stack model can predict the temperatures and voltages accurately compared with the original stack model. At the same time, the time and memory required for computation are reduced by approximately 60% for a short stack model containing seven cells, owing to the simplified fuel-side mass transfer and charge transfer processes. For a large stack model with 21 cells, the reduction in computation time can even exceed 70%. The reduced computational cost makes it possible to simulate the models of industrial-scale FCSOC stacks with up to 61 cells.

NetKernel: Making Network Stack Part of the Virtualized Infrastructure

This paper presents a system called NetKernel that decouples the network stack from the guest virtual machine and offers it as an independent module. NetKernel represents a new paradigm where network stack can be managed as part of the virtualized infrastructure. It provides important efficiency benefits: By gaining control and visibility of the network stack, operators can perform network management more directly and flexibly, such as multiplexing VMs running different applications to the same network stack module to save CPU cores, and enforcing fair bandwidth sharing. Users also benefit from the simplified stack deployment and better performance: For example mTCP can be deployed without API change to support nginx natively, and shared memory networking can be readily enabled to improve performance of colocated VMs. Testbed evaluation using 100G NICs shows that NetKernel preserves the performance and scalability of both kernel and userspace network stacks, and provides the same isolation as the current architecture.

Highly efficient nondoped bilayer organic light-emitting diodes based on triphenyl phosphine oxide protected iridium complexes

Nondoped phosphorescent organic light-emitting diodes (OLEDs) with simplified stacks are desired for practical displaying and lighting applications. However, doping emissive layers are commonly adopted due to serious triplet quenching of phosphors. Here, we demonstrate high-efficiency nondoped bilayer OLEDs based on triphenylphosphine oxide protected iridium(III) complexes. The host feature of peripheral phosphine oxide groups effectively suppresses intermolecular interaction induced quenching. As consequence, the maximum external quantum efficiency beyond 20% and near-zero roll-offs at 1000 nits were achieved, comparable to the best nondoped phosphorescence devices.

Significantly Improving the Durability of Single-Chamber Solid Oxide Fuel Cells: A Highly Active CO2-Resistant Perovskite Cathode

The single-chamber solid oxide fuel cell (SC-SOFC) represents an attractive alternative to batteries for micropower applications given its simplified stack design, easy gas management, lower operation temperature, and quicker start-up. Its stable operation over long duration relies on the availability of CO2-resistant cathode materials with high oxygen reduction reaction (ORR) activity. Here, we reported a novel perovskite oxide cathode SrSc0.075Ta0.025Fe0.9O3−δ (SSTF75), which provides CO2 resistance and high ORR activity. A peak power density of 1430 mW cm–2 was achieved with the SSTF75/Sm0.2Ce0.8O1.9 cathode at 650 °C with a methane and oxygen gas mixture (1:1 v/v ratio). It retained stable voltage performance at 600 °C for an over 90 h duration when discharged at a current density of 1000 mA cm–2. This work not only represents an advancing step in the SC-SOFC field but also authenticates an effective codoping strategy for the design of a CO2-resistant cathode for low-temperature applications.

Optimizing the Optical and Thermal Design of Heat-Assisted Magnetic Recording Media

An optimization study of heat-assisted magnetic recording (HAMR) media optical and thermal designs is presented. A simplified stack with five component layers is used, which allows a systematic exploration of their optical and thermal properties parameter space. The optimum HAMR media design maximizes the downtrack temperature gradient at a set target write width and for a laser power limited by the near-field transducer (NFT) lifetime. Best absorption results are obtained for plasmonic-like underlayer/heat sink and a recording layer with low refractive index and low extinction coefficient. This combination of optical properties leads to the optimum confinement of the electromagnetic field between the NFT and the lower interface of the recording layer. The optimal optical design is not sensitive to the three NFT designs considered and to the thermal optimization. For the thermal properties optimization, weighted gradient and power are introduced that account for the variations of recording velocity across the disk surface in 7200 rpm hard disk drives. The ideal thermal design contains a strong heat sink with an optimal effective thermal conductance heff between the recording layer and the heat sink. Sufficient thermal anisotropy is required in the recording layer to minimize the lateral heat flow in that layer. The optimum heff also depend on disk velocity and in-plane thermal conductivity in the recording layer. This paper also present the quantitative comparisons of different optical and thermal tradeoffs/variations that might be necessary in practical HAMR media designs.

Investigation of the flow field inside the manifold of a real operated fuel cell stack using optical measurements and Computational Fluid Mechanics

The versatility of fuel cells enables a wide range of applications. Usually fuel cells are combined to stacks such that the reactant supply of the single cells is achieved via a pipe branching system, the manifold. The overall performance significantly depends on cell flow rates which are related to the fluidic interaction of the manifold and the cells. Computational Fluid Dynamics (CFD) simulations, which are often used to find a suitable design, lack experimental flow data for validation of the numerical results. To enable flow measurements within the small geometries of the manifold and to provide reliable velocity information inside a real fuel cell stack, a low-coherence Laser Doppler Anemometer (LDA) is applied, which uses multi-mode laser light to achieve a spatial resolution of <100 μm. The use of fluorescent particles and backward scatter mode make the sensor highly suitable for the application in small manifold geometries like in fuel cell stacks. Sensor and measurement technique are validated in simplified stack models and the applicability to air flows is demonstrated. Finally, for the first time, velocity profiles with high spatial resolution inside an operated fuel cell stack are presented, which serve as benchmark for CFD to find an optimal geometry.

Nanostructure Electrodes for Metal-Supported Solid Oxide Fuel Cells

Metal-supported solid oxide fuel cells (MS-SOFCs) use thick porous alloy substrates as the mechanical supports for thin functional anode-electrolyte-cathode layers and have gained increasing interest due to their advantages over the traditional all-ceramic counterparts such as increased structural robustness, improved tolerance toward rapid thermal cycling, simplified stack assembling and reduced materials costs as well. However, issues such as coarsening of the active nickel anode catalysts, metalic interdiffusion between the alloy substrates and the anodes, and application of appropriate cathode layers remain to be addressed. Here we report some of recent progress in nano-scale electrode catalysts for MS-SOFCs using the liquid phase impregnation technique.

Novel Bumping and Underfill Technologies for 3D IC Integration

In previous work, novel maskless bumping and no-flow underfill technologies for three-dimensional (3D) integrated circuit (IC) integration were developed. The bumping material, solder bump maker (SBM) composed of resin and solder powder, is designed to form low-volume solder bumps on a through silicon via (TSV) chip for the 3D IC integration through the conventional reflow process. To obtain the optimized volume of solder bumps using the SBM, the effect of the volumetric mixing ratio of resin and solder powder is studied in this paper. A no-flow underfill material named “fluxing underfill” is proposed for a simplified stacking process for the 3D IC integration. It can remove the oxide layer on solder bumps like flux and play a role of an underfill after the stacking process. The bumping process and the stacking process using the SBM and the fluxing underfill, respectively, for the TSV chips are carefully designed so that two-tier stacked TSV chips are sucessfully stacked.

Zn(O, S) layers for chalcoyprite solar cells sputtered from a single target

A simplified Cu(In, Ga)(S, Se)2 /Zn(O, S)/ZnO:Al stack for chalcopyrite thin-film solar cells is proposed. In this stack the Zn(O, S) layer combines the roles of the traditional CdS buffer and undoped ZnO layers. It will be shown that Zn(O, S) films can be sputtered in argon atmosphere from a single mixed target without substrate heating. The photovoltaic performance of the simplified stack matches that of the conventional approach. Replacing the ZnO target with a ZnO/ZnS target may therefore be sufficient to omit the CdS buffer layer and avoid the associated complexity, safety and recycling issues, and to lower production cost.

Low-driving-voltage and colour-stable white organic light-emitting diodes with a cross-patterned multi-emissive layer

We have applied a simple cross-patterning technique for the fabrication of phosphorescent white organic light-emitting diodes (WOLEDs) with red, green and blue (RGB) emitters; the resulting device has relatively low driving voltage and high colour stability. The selectively cross-patterned multicolour emitting layer (EML) was easily prepared using a metal mask without an alignment process. Not only was the characteristic of low driving voltage obtained but also of improved colour stability, which can be ascribed to the simplified stack of the EML and the corresponding suppression of the biased shift in the recombination zone. The spatial distribution and variation of the stacked EML structure could explain the origin of the robust white emission. Compared with the conventional WOLED with a RGB simple stack, the cross-patterned multi-EML device showed a slight change in colour coordinates in the luminance range 100–8000 cd m−2, with a decrease in the driving voltage of 0.5–2.0 V, while the luminous efficiency was maintained.

A Model-Based Approach to the Common-Diffraction-Surface Stack Method – A Synthetic Case Study

The Common-Reflection-Surface (CRS) stack method parameterizes and stacks seismic reflection events in a generalized stacking velocity analysis. It considers a discrete number of events contributing to a given stack sample such that conflicting dip situations can be handled. The reliable detection of such situations is difficult and missed contributions to the stacked section cause artifacts in a subsequent poststack migration. This is deleterious for complex data where prestack migration is no viable option due to its demands on velocity model accuracy, such that we might have to rely on poststack migration. As an alternative, the conflicting dip problem has been addressed by explicitly considering a virtually continuous range of dips with a simplified stacking operator in a process termed Common-Diffraction-Surface (CDS) stack. In analogy to the CRS stack, the CDS stack has been implemented and successfully applied in a data-driven manner based on coherence analysis in the prestack data. In view of the computational costs, we present a more efficient model-based approach to the CDS stack designed to generate stack sections optimized to image discontinuities by poststack migration. This approach only requires a smooth macro-velocity model of minor accuracy. We present our results for the synthetic Sigsbee 2A data and compare them to the results of CRS stack and datadriven CDS stack.

Use of Nucleic Acid Analogs for the Study of Nucleic Acid Interactions

Unnatural nucleosides have been explored to expand the properties and the applications of oligonucleotides. This paper briefly summarizes nucleic acid analogs in which the base is modified or replaced by an unnatural stacking group for the study of nucleic acid interactions. We also describe the nucleoside analogs of a base pair-mimic structure that we have examined. Although the base pair-mimic nucleosides possess a simplified stacking moiety of a phenyl or naphthyl group, they can be used as a structural analog of Watson-Crick base pairs. Remarkably, they can adopt two different conformations responding to their interaction energies, and one of them is the stacking conformation of the nonpolar aromatic group causing the site-selective flipping of the opposite base in a DNA double helix. The base pair-mimic nucleosides can be used to study the mechanism responsible for the base stacking and the flipping of bases out of a nucleic acid duplex.

Experimental evaluation of an ambient forced-feed air-supply PEM fuel cell

Increasing interest in utilizing proton exchange membrane fuel cells for sub-kilowatt mobile applications has prompted the need for fundamental understanding of operating characteristics of polymer electrolyte membrane (PEM) fuel cells. However, there is a few published data regarding related information, especially for PEM fuel cells with ambient force-feed air-supply and their stack design which is the subject of this study. Although the air for the cathode was directed from ambient atmosphere with variable humidity, the flow rates were controlled in order to simulate their effects on the cell performance under real conditions. In addition, the steady-state performance and transient response for H2/air PEM fuel cells were investigated under a variety of load cycles and operating conditions. Impact of H2 humidity on the performance was negative for this simplified stack design under appropriate operation current output and the dry H2 was utilized for the rest of this study. It was found that the humidity of cathode inlet gas had a significant effect on fuel cell performance. When the air relative humidity was higher than 55%, the stack operation resulted in more stable and higher performance. In addition, high moving rate of fuel cell stack (i.e. 20.3kmh−1) was found necessary for supplying air to the fuel cell directly and indirectly in order to prevent stack from over-heating.

The effect of fuel cell operating conditions on DIR-MCFC system selection

The use of natural gas and ‘direct internal reforming’ allows very simple molten carbonate fuel cell systems to be devised. The compositions and flow rates of the anode and cathode stack inlet and outlet gas flows can be described by five dimensionless system variables. These variables describe the total air flow, the cathode recycle, the overall fuel utilisation, the anode recycle, and the amount of steam addition. The equilibrium stability limits of carbon are presented and related to the system variables. A simplified stack model has been developed and a parametric analysis has been performed to determine the effect of the system variables on the stack performance. Two atmospheric pressure systems are presented which are suitable for combined heat and power applications of less than 1 MWe.